The primary purpose of the proposed series of investigations is to gain better understanding of the antidepressant effects of sleep deprivation by analyzing polysomnography (PSG) before and after a night of PSD in depressed patients compared to normal controls. The secondary aim of the project involves analyzing SPECT cerebral blood flow (CBF) data in these subjects for correlations with clinical changes. The investigator predicts: 1) depressed responders will show increased rapid eye movement latency, increased slow wave sleep, and increased sleep efficiency following PSD; 2) responders' abnormal CBF distribution will become less abnormal after PSD and 3) the magnitude of such changes will correlate with the intensity of clinical response. Subjects will consist of physically healthy men and women diagnosed with major depression (unipolar) with baseline 17-item Hamilton Depression Rating Scale (HDRS-17) > 16. Normal controls will be matched for age and gender. Exclusion criteria include other psychiatric disorders; pregnancy; conditions which might be exacerbated by SD; sleep disorders; irregular sleep cycle; neurological or circulatory conditions which could affect electroencephalography (EEG) or SPECT; and recent use of substances which could affect sleep pattern, EEG, and/or SPECT. Subjects will undergo an adaptation night, a baseline night, a night of PSD, and a "recovery" night in the sleep laboratory with standard recording montage. Data will be analyzed with the Delta Pass system (period-amplitude analysis.) Subjects will undergo cerebral SPECT with HMPAO during baseline and PSD afternoons. Subjective and clinician-administered mood and depression rating scales will be administered at standard times during baseline and recovery days and during the PSD night. The study, a 3 X 2 mixed design, has 3 groups (depressed responders, depressed nonresponders, and controls) and 2 conditions (baseline and following PSD.) A drop in modified HDRS-17 (e.g., omitting items 4-6 (sleep) and 16 (weight loss) > 30 percent following PSD will be used as a criterion for clinical response. PSG and SPECT data will be analyzed by multivariance analysis of variance.